Foundation models focused on text-to-x are quite ubiquitous now and for a good reason. Even thought-to-x is emerging (Nubrain, Condu.it). It’s becoming more and more obvious why. We are better at building models that can digest already digitized content (text, waves).
For me, smell-to-x is translunar and massively underrated.
Here is why: It is generally known that scientific studies on olfaction are underfunded and far less promising in comparison to ML approaches in other areas where it is relatively easier to digitize data from either existing or “easy to harvest” datasets. Moreover, where it is easier to envision the real-world application of text-to-x, it’s more difficult to envision the monetization of smell-to-x.
Using its own set of molecules (matter), the team has developed a foundation model that uses these molecules to create a photograph of scent.
The business model:
- Replicate expensive and carbon-intensive Rose Bulgaria oil by matching its receptor activation pattern with more sustainable or cost-effective molecules
- Unlock opportunities for safer and more captivating blockbuster fragrance molecules.
- Develop molecules for olfactory receptors in the skin that have been found to imbue therapeutic properties like anti-aging and wound healing.
Patina is stirring up a massive market with the foresight for commercializing something that has never been sold quite like this before.
